Functionalism and the Meaning of Social Facts

FUNCTIONALISM AND THE MEANING OF SOCIAL FACTS

Warren Schmaus

Abstract

This paper defends a social functionalist interpretation, modeled on psychologicalfunctionalism, of the meanings of social facts. Social functionalism provides a better explanationof the possibility of interpreting other cultures than approaches that identify the meanings of socialfacts with either mental states or behavior. I support this claim through a functionalistreinterpretation of sociological accounts of the categories that identify them with their collectiverepresentations. Taking the category of causality as my example, I show that if we define it insteadin terms of its functional relations to moral rules, it becomes easier to recognize in other cultures.

1

FUNCTIONALISM AND THE MEANING OF SOCIAL FACTS

A functionalist approach to the meaning of social and cultural facts that is analogous tofunctionalism in the philosophy of mind makes it easier to explain how it is possible to interpretother cultures. This social functionalism provides a better explanation of the meaning of socialfacts than do attempts to reduce them to either behavior or the mental states of the members of thesociety in question. Also, it does not leave their meaning dependent on the interpretations ofethnographers. It defines social facts in terms of their functional relationships to other social facts,environmental conditions, and types of actions. Much as psychological functionalism emphasizesthat the same type of mental state can be instantiated in multiple types of brain states, socialfunctionalism emphasizes that a type of social fact can be instantiated in multiple types of mentalstates. Hence, somewhat ironically, where the relative strength of psychological functionalismover behaviorism and materialism derives from its appeal to the content of mental states, thestrength of social functionalism comes from keeping such mental contents at arm's length. I willillustrate the strength of this approach through a functional re-interpretation of Durkheimiansociology of knowledge. Specifically, I will show that identifying the concept of causality with itssocial functions rather than with its collective or cultural representations makes it easier torecognize this concept in other cultures.

Durkheim originally conceived collective representations as a type of mental entity sharedby the members of a society. He included among a culture's collective representations not only itsreligious and moral ideas but also its categories of causality, substance, space, and time. Forinstance, Durkheim and Mauss (1903) identified the Zuñi category of space with their collectiverepresentation of the division of space into seven regions named for the seven clans in their tribe. On my account, it is not necessary for the Zuñi all to have the same type of mental representation inorder for them to understand this division of space. It is sufficient that they are able to participate

in social functions that require their being able to specify and agree upon locations. Similarly, tosay that the Chinese traditionally conceived time as cycles of yin and yang is not to say that all themembers of this society must have the same representation of time. The meaning of the division ofthe year into periods of yin and yang has to do with such things as the organization of agriculturaland domestic labor (Granet 1922). Also, although social life depends on moral rules that assumethat people are causally responsible for their actions, it is not necessary that everyone in the samesociety represent the concept of causality to himself or herself in the same way. What is important,however, is that they are able to agree on assignments of moral responsibility.

I do not mean to suggest that social scientists still adhere to Durkheim's mentalist sense ofthe term \"collective representation,\" or even that there is a univocal meaning of this term in thesocial sciences, for indeed there is not. Radcliffe-Brown (1952), for instance, regarded collectiverepresentations as explicit, public representations such as myths, cosmologies, ritual statements,symbols, concrete images, artifacts, or gestures. Yet some who identified collective with publicrepresentations thought that we could infer how people in other cultures thought from an analysisof these representations. Indeed, the assumption that the analysis of myths and cosmological ideassheds light on how people think goes back to Tylor and Frazer, and continues through Durkheim,Lévy-Bruhl, and Lévi-Strauss. Although this assumption has lately come under increasing attack(Bloch 1977:290; Boyer 1994:112; Cole and Scribner 1974:143; Holy and Stuchlik 1983:100ff;Jahoda 1982:219), some of its critics nevertheless continue to locate culture and meaning in theminds of individual members of society. To cite but one of many possible instances, Holy andStuchlik hold that \"the term `culture' refers, to put it bluntly, to what is in people's heads, to theknowledge they have\" (1983:21).

Geertz rejects the mentalistic conception of culture, arguing that meaning and hence cultureis public (1973:12). Public expressions of one's culture, however, far from defining social facts,are among the very social facts that need to be interpreted. Geertz makes this point through the useof the distinction between \"thin\" and \"thick\" descriptions of behavior. He discusses the example

1

of distinguishing mere twitches of the eye from winks, parodies of winks, practice winks, andfake winks. All share the same thin description of a rapid contraction of the eyelid. What separatethem are thick descriptions of their meanings in some public, social context (1973:6-7). A thickdescription is a construction imposed on a social action by an anthropologist and not necessarilythe meaning that it has for the agent herself. The most important issue for appraising such aninterpretation, he says, is how well it sorts things into kinds, how well, for example, it sorts \"realwinks from mimicked ones.\" For Geertz, one sorting appears to be better than another just to theextent that it brings us into closer \"touch with the lives of strangers\" (1973:15-16). Presumably,then, he does not think that we can impose any arbitrary interpretation on a culture: there are realworld constraints on the way we sort cultural items.

However, Geertz offers the reader no clue as to how we know when an interpretation hassorted actions into real and not just fictional kinds, or indeed what separates social actions into realkinds if it is not the meanings they have for their agents. Perhaps, then, a functionalist approach tothe meaning of social facts will at least provide us with a heuristic for sorting actions into kinds, aheuristic that depends on their meaning neither for their agents nor for their ethnographicinterpreter.

Social and Psychological FunctionalismAs I indicated above, the functionalist approach to the meaning of social facts that Iadvocate derives by way of analogy from psychological functionalism. It does not derive from thefunctionalism of Malinowski or Parsons or include any hypotheses about the functional unity of asociety or culture. To try to avoid confusion with older sociological functionalisms, I will use theterm \"social\" functionalism as a name for my approach to the meaning of social facts.

Psychological functionalism, as the members of this audience well know, does not identify types

of mental states with either types of brain states or types of behavior. Rather, it defines them interms of their functional relationships to other mental states, sensory inputs, and behavioraloutputs. Similarly, my social functionalism does not identify types of social facts with types ofrepresentations, whether mental or public. Instead, it defines types of social facts in terms of theirfunctional relationships to other social facts, environmental inputs, and behavioral outputs. Twofacts are of the same type when they bear the same such relationships.

Psychological functionalism allows for multiple physical instantiations of the samefunctionally defined type of mental state. Whether a group of mental phenomena constitutes a typeof mental state is determined by whether some psychological generalization holds true of them. Onthe other hand, whether some group of neural phenomena constitutes a type of brain state isdetermined by whether some neurophysiological generalization holds true of them. Even if everyindividual mental state is an individual brain state, there is no reason to believe that the laws ofpsychology and neurophysiology will divide these states into the same classes. Two people whoshare the same type of mental state, defined in terms of its psychological function, do not therebyshare the same type of brain state.

The earliest functionalist theories in the philosophy of mind tended to explain the multipleinstantiability thesis through analogies with computers and other machines (e.g., Putnam 1967;Block and Fodor 1972). A common objection to this machine functionalism was that in definingtypes of mental states purely in terms of their functional relations, it would allow computers andother devices as well as minds to have mental states (e.g. Block 1978). This objection, I think,rests on an equivocation regarding the notion of function. When we talk about the functionalrelationships among mental states, we are using \"function\" in the sense we invoke when weexplain the existence and structure of something in terms of the purpose it serves. In livingorganisms, not all the causal relations a structure has are functional relations, but only those wherethe effect somehow feeds back to maintain the organism and thus the structure in existence. Wemay try to write a program that represents all the functional relationships among our mental states,

1

but the relationship between one step in the program and another would be functional, strictlyspeaking, only in the mathematical sense of the term. When we run the program on a computer,these mathematical relationships are physically realized as causal relationships among machinestates. We only ascribe purposes to computers that are not really their own. Thus, it would beentirely out of place to provide a functional explanation, in the psychological or biological sense, ofthe state of a computer. It is difficult to see how a mind and a computer could then be said to be inthe same type of state if we could not explain that state in the same way for each.

We can then generalize the multiple instantiability thesis to include the relationship betweensocial facts or institutions and their psychological realizations. Members of a society may sharemoral rules, religious beliefs, and other concepts that are all the same from the point of view oftheir social functions. However, there is no reason to believe that all the individual members ofthis society thereby have the same kinds of mental states, defined in terms of their psychologicalfunctions. Psychological functions are distinct from social functions. Also, just as the same typeof brain state may instantiate different psychological functions in different individuals, the sametype of mental state can instantiate different social functions. There is a many-to-many relationshipbetween types of social facts and types of mental states, just as there is between types of mentalstates and types of brain states. Hence, there seems to be little reason for the social sciences toinvoke the notion of collective representations, when these are understood as shared mental states.

Sperber might object that in allowing for only token-token identities between social factsand psychological states, my social functionalism undermines the possibility of naturalizing thesocial sciences. As he sees it, laws or generalizations in the social sciences must be about types,not tokens (1996:6). Hence, he opts for a \"materialist\" ontology of mental representations, whichhe regards as \"brain states described in functional terms,\" and the causal chains that connect them(1996:26-27). In reply, I would argue that there is no reason that functional types at one level ofexplanation must be the same as those at another. If there were any merit to Sperber's objection, itwould seem to have to be valid all the way down. That is, there could then be no generalizations

about organs, tissues, cells, cell organelles, and so on unless they could be expressed in terms ofthe natural kinds of physics, which is absurd, as functional kinds do not even exist at that level. Furthermore, Sperber's approach presents an obstacle to cross-cultural interpretation. ForSperber, one's mental representation is of the same type as another person's only if it belongs tothe same causal chain. Presumably, when an ethnographer encounters an entirely new culture,none of the representations in that culture will then belong to the same causal chains as theethnographer's. This would then lead to a radical conceptual relativism, in which the ethnographerwould be unable to say that that culture had any representations of the same type as hers, includingany representations of causality, time, permanent substance, space or place, and so on.

The multiple instantiability thesis still holds if we re-interpret collective representations aspublic representations or behavior. That is, the same type of social fact, functionally defined, mayhave more than one sort of public representation. Also, the same behaviors, words, and symbolsmay have different meanings in different social contexts. There are multiple behavioral correlatesfor any meaning and the same observable behavior can have many different meanings.

Social functionalism has advantages over behaviorism in so far as it provides a way ofarticulating the contextual meaning of behavior by bringing out its systematic relationships withother social facts. To identify the presence of a moral norm with the expression of indignation atits violation, as behaviorist sociologists are wont to do, is much too crude. Even putting aside theproblem of the presence of moral rules when they are not being violated, sociological behaviorismfails to distinguish moral rules properly so-called from other kinds of norms. Concertgoers mayexpress indignation at the performance of a new work of music and sports fans may be outragedby an umpire's call. I do not wish to deny that there may be some similarities among these kindsof situations. Nevertheless, they are different and behaviorism fails to bring out these differences. It is not even clear that behaviorism allows one to distinguish indignation from other sorts ofanger. The very terms \"indignation\" and \"outrage\" carry the connotation that there are reasons forthe expression of anger connected with the violation of norms, rules, or expectations, whether

1

these be moral, artistic, epistemic, or athletic. The mere observing of anger will not indicatewhether there are such reasons.

To distinguish indignation from anger and to distinguish the different forms indignationmay take, we must appeal to the meanings of these expressions of emotion. Of course, theirmeanings for the individual social actors who express these emotions may include their consciousawareness of their mental representations of these meanings. However, the meaning of the actionfor the individual may not be the same as its social functional meaning. Furthermore, even if it isthe same, this social functional meaning may be represented in more than one way. For instance,there is no reason to assume that all the social actors expressing indignation at the violation of anorm will all represent this norm to themselves in the same way.

One may think that what I am arguing for is merely a functionalist reinterpretation of thenotion of collective representations and that I would be over-stating my case to claim that the socialsciences may dispense with collective representations entirely. I want to resist this interpretation ofwhat I am doing. If the meaning of a social fact is just a node in a network of functional

relationships among social facts, why should we consider it a representation at all? What does thisnode represent? How does it represent? To whom does it represent? I do not see answers to suchquestions as readily forthcoming and I see no reason to trouble ourselves with them or to hold outfor answers to them.

It may help to recall that for Durkheim, collective representations were states of thecollective consciousness. Because it opened him up to the group mind interpretation and objection,Durkheim dropped the term \"collective consciousness\" midway through his career and left thequestion of to whom collective representations are present an open question. This questioncontinues to go unanswered by many contemporary social scientists, who postulate collectiverepresentations in order to give rational interpretations of what would otherwise appear to beirrational behavior. As Boyer explains, these collective representations \"do not describe thoughtsthat occur to actual people; they describe thoughts that people might entertain, in the

anthropologist's view, if they wanted to make sense of what they actually do and say\" (1994:51). Why then call these notions collective representations? I do not mean to deny that a social scientistmay represent to herself a network of functional relationships, but then this would be an individualand not a collective representation.Functionalism and the Sociology of KnowledgeThe sociology of knowledge may be the social science that stands most in need of afunctionalist re-interpretation. The identification of the categories of causality, substance, space,and time with their collective representations is an important premise in the arguments of those whomaintain that reality is a culturally variable construction. This position, if true, would rule out anypossibility of interpreting other cultures, for how could one make sense of the actions of peoplewho lived in a different reality? Of course, this constructivist conclusion also depends onadditional premises. These include the assumption that these categories shape our perception of theworld, that collective representations depend on social causes and thus are culturally variable, andthat the individual mind comes into the world as a blank slate and passively acquires a set ofcollective representations from her culture. Of course, as Tooby and Cosmides (1989), Cole andScribner (1974), and Hallpike (1979) have argued, there is absolutely no basis in experimentalpsychology for the last assumption. However, the whole constructivist edifice crashes to theground once we remove the premise that the categories of the understanding are collectiverepresentations.

To identify the meanings of the categories with their social functions, on the other hand,would help us to explain how communication with and interpretation of other cultures is possible. Consider, for example, a functionalist interpretation of the category of causality that identifies thisconcept in other cultures through its relations to moral and legal rules. Human society as we knowit would not be possible without such rules. To have rules, people must be held accountable for

1

their actions, but that assumes that in some sense they are the cause of their actions. Hence, allsocieties must have some concept or other of causality. Some may object that human societiesexercise social control through sanctions or the threat of sanctions. Sanctions, however, areapplied as the result of the violation of a rule. Also, even the threat of sanctions is not alwaysimmediately present and in their absence most people nevertheless continue to follow the rules. Ofcourse, many of these rules may be only implicitly understood and not carefully articulated in alegal or moral code. Nevertheless it would not be possible to have even implicit moral ruleswithout some notion of responsibility.

Durkheim, in fact, argued that our concept of a causal relation originated from our notion ofmoral obligation. For Durkheim, a causal relation is a necessary relation. He held that the notionof a necessary relation derived from the obligation of members of society to participate in religiousrites. In certain rites exemplified by indigenous Australians, for instance, one imitates a certainspecies of plant or animal at an appropriate time of year in order to make it reproduce and flourish. Society imposes the obligation to imitate this species because a social interest is at stake. Toobligate the members of a society to imitate an animal or plant so that it will reproduce is topresume that performing the rite necessarily leads to the flourishing of the species that is beingimitated. If society allowed people to doubt this causal relationship, Durkheim argued, it could notcompel them to perform the rite (1912:524ff [1995:370ff]). To be sure, in this example Durkheimmay have been less than clear about whether he was providing an account of the causal origins orthe function of the concept of causality. However, I think we can separate out a functional accountthat would include the premise that society cannot obligate someone to do something without someconcept of causality.

Similarly, Lévy-Bruhl described a notion of participation that plays the functional role ofcausality in so-called \"primitive\" societies. In accordance with this notion of participation, peopleare held responsible for all sorts of things for which we would not blame them. For example,according to Lévy-Bruhl, for the primitive there is no such thing as an accidental death or death by

disease or other natural causes. All death is due to witchcraft. Witchcraft assumes a notion ofparticipation, according to which one is supposed to be able to harm one's intended victim throughactions taken against his or her bodily fluids, hair, nails, footsteps, image, articles of clothing,utensils, etc. because all these things \"participate\" in the victim. People who perform suchwitchcraft may be held responsible for the death of their victims (Lévy-Bruhl 1910:321ff[1985:276ff]; 1922:20ff [1978:37ff]; 1927 [1928:114ff]). Although we may not hold people toaccount for murder through witchcraft, nevertheless the relation of the notion of participation tomoral responsibility allows us to recognize participation as a causal concept. Boyer, however,questions whether different cultures actually do have different concepts of causation and assertsthat \"people do not plow their fields . . . in terms of `participation'\" (1994:129). To the extent thathe is right, however, this fact merely shows that there may be more than one concept of causalityoperating even in one and the same so-called \"primitive\" society, a point that Lévy-Bruhl wouldhave readily conceded (1922:517 [1978:442-43]).

That these various peoples cited by Durkheim and Lévy-Bruhl may have been mistakenabout these particular causal relationships is beside the point. The point of these examples is thatthe category of causality is necessary for imposing moral obligations. By interpreting the categoryof causality in terms of its social functional role, I do not mean to suggest that the concept ofcausality is the same for all societies. My everyday notion of causality, for instance, has nothing todo with morally culpable death caused by witchcraft or with any obligation to imitate totemicspecies. Many in our society may conceive causality as a statistical rather than a necessaryrelationship between cause and effect. There may even be functional explanations of the culturaldifferences among concepts of causality that appeal to the specific roles these concepts play in eachsociety. Nevertheless, it is their functional relationships to moral rules that bring various conceptsunder the category of causality and thus that allow for the cross-cultural interpretation of conceptslike participation as causal concepts.

Of course, a society may impose obligations on its members that may not be necessary for

1

its continued survival. This concession, however, does not stand in the way of providing afunctional account of the category of causality. The category of causality may be presupposed bysuperfluous rules as well as those rules necessary for the maintenance of the society in question. However, the society is better off having these extra rules than having no concept of causality andthus no rules at all. The persistence of societies depends on moral rules and thus the category ofcausality helps to maintain the society.

Indeed, even for someone to think that he or she is not obligated to perform a rite because itdoes not work or to deny responsibility for witchcraft would involve a negative use of the categoryof causality. The notion of working suggests that we may want to generalize our functionalaccount of the category of causality to include its relations to rules of technique and hygiene as wellas moral rules. In fact, Lévy-Bruhl suggested that in non-literate societies, it might be difficult todistinguish purely technical procedures from moral and religious rules. For example, he describedhow in making an ax, the primitive must first propitiate certain spirits before cutting down the treefrom which he will obtain the wood for the handle (1928:26).

My functional account of the category of causality should be understood not merely assome sort of transcendental argument regarding the necessary conditions for social life or for itsinterpretation. Rather, it can also be read as reasoning to an empirically testable hypothesis. Questions as to which categories are functional and which not and what social functions they serveare topics for empirical investigation. One way to test the claim that the category of causality hasthe function of holding society together by making moral and other rules possible is to see whetherin fact the category is universal, that is, whether in fact all societies have some concept or other ofcausality. In testing this claim, as Boyer (1994:112) points out, we must be careful not to relysolely on ethnographic accounts of the myths, cosmologies, and religious beliefs of variouscultures. Typically, mythologies deliberately violate people's intuitive expectations about eventsand their causes. The concepts that people in various societies actually have, Boyer argues, shouldbe elicited through the same kind of cognitive psychological experiments that we would use to

investigate how people in our own society think (1994:291-92). Such experiments could also beused to determine whether certain categories are even universal within our own society, or whetherthey are lacking in certain pathological cases in which individuals have difficulty with socialrelationships.

Some may object that experimental subjects, finding the cognitive science testing situationhighly artificial, will provide responses that do not reflect their true thinking (e.g. Holy andStuchlik 1983, ch. 4). Subjects from non-Western societies especially tend to give either theanswers that they think are expected of them or those that will quickly bring to a close discussionsof topics they are not interested in pursuing. This problem, however, strikes me as a technical andmethodological difficulty rather than an insuperable barrier for this program of research. Furthermore, the results of cognitive experiments could be controlled by investigations intowhether there are corresponding syntactic categories in the language of the culture in question. Ofcourse, from the facts that the syntactic categories make language possible and that language makeshuman social life possible, it does not follow that all of the syntactic categories have socialfunctions. Gender in modern European languages is a notable counter-example. Hence, theanalysis of languages alone could not reveal the social functions of the categories but must bepursued in conjunction with cross-cultural studies of cognition.

However, even if a category could be shown to be universal among human cultures, suchevidence would not suffice to establish that it had a social function. We need some way todistinguish genuine functional accounts from cases in which it would merely appear that having acertain concept benefits society. There must be some account of how these purported benefits helpmaintain the society in existence. If we give a natural selectionist account, one might argue, itwould be necessary to assume that this beneficial concept was actually represented in the individualminds of the members of the society that it benefited. For example, I have been arguing that one ofthe functions of the category of causality is to make moral rules possible. In order for theindividual members of society to follow these rules, they must be able to represent the concept of

1

causality to themselves. What this objection shows, however, is that individual mental

representations may play a role in mediating functional explanations of the categories. It in no wayestablishes a role for collective representations. Natural selection may favor an ability to representcausal concepts without selecting for any particular causal concept or for members of a socialgroup to represent this concept in the same way.

ConclusionIn conclusion, the meanings of social facts should be understood in terms of theirfunctional relationships with other social facts, environmental conditions, and behavioral outputs.Social functionalism relates the meanings of social facts to behavior without reducing them tobehavior. Allowing for the instantiation of social facts in multiple types of psychological states,social functionalism dispenses with the notion of collective mental representations. Individualmental representations, however, may still play a role in explaining the relationship between theindividual and society, in mediating functional explanations, and in explaining individual actions.

BIBLIOGRAPHY

Bloch, M. (1977), \"The Past and the Present in the Future\Block, N. (1978), \"Troubles with Functionalism\Cognition: Issues in the Foundations of Psychology: Minnesota Studies in the Philosophyof Science, vol. 9. Minneapolis: University of Minnesota Press, pp. 261-325.Block, N., and Fodor, J. (1972), \"What Psychological States are not\81:159-82.

Boyer, P. (1994), The Naturalness of Religious Ideas. Berkeley: University of California Press.Cole, M., and S. Scribner (1974), Culture and Thought. New York: John Wiley & Sons, Inc.Durkheim, E. (1912), Les Formes élémentaires de la vie religieuse. Paris: Alcan. Trans. by K.Fields as The Elementary Forms of Religious Life. New York: Free Press, 1995. Durkheim, E. and M. Mauss (1903), \"De quelque formes primitives de classification: contribution

à l'étude des représentations collectives\Needham as Primitive Classification. Chicago: University of Chicago Press, 1963.Geertz, C. (1973), The Interpretation of Cultures. New York: Basic Books.Granet, M. (1922), The Religion of the Chinese People, trans. by M. Freedman . New York: Harper Torchbooks, 1975.

Hallpike, C. R. (1979), The Foundations of Primitive Thought. Oxford: Clarendon Press.Holy, L., and M. Stuchlik (1983), Actions, Norms, and Representations. Cambridge: CambridgeUniversity Press.

Jahoda, G. (1982), Psychology and Anthropology: a psychological perspective. London: Academic Press.

Lévy-Bruhl, L. (1910), Les fonctions mentales dans les sociétés inférieures. Paris: Alcan. Trans.in 1926 by L. A. Clare as How Natives Think. Princeton: Princeton University Press,14

1985.

Lévy-Bruhl, L. (1922), La mentalité primitive. Paris: Alcan. Trans. in 1923 by L. A. Clare asPrimitive Mentality. New York: Macmillan, 1978.Lévy-Bruhl, L. (1927), L'Ame primitive. Paris: PUF, 1963. Trans. in 1928 by L. A. Clare asThe Soul of the Primitive. New York: MacMillan.Putnam, H. (1967), \"Psychological Predicates\and Religion. Pittsburgh: University of Pittsburgh Press.Radcliffe-Brown, A. R. (1952), Structure and Function in Primitive Society. London: Cohen &West, Ltd.

Sperber, D. (1996), Explaining Culture: a naturalistic approach. Oxford: Basil Blackwell, Ltd.Tooby, J., and L. Cosmides (1989), \"Evolutionary Psychology and the Generation of Culture,

Part I: Theoretical Considerations\